American Sign Language Recognition using Convolutional Neural Network
Sadaf Ikram, Namrata Dhanda
Abstract
There have been a lot of research and advancements for helping people who cannot speak and listen with the help of modern technology. People like that can't even utilize voice applications like Google assistant, Siri etc. American Sign Language (ASL) is such complete language which contains various signs using hand movements for dumb & deaf people. The main focus of this paper is on using Deep Learning methodologies to analyze the sign language symbols and make useful predictions accordingly. Sign language is not known by many people so it will help the people who are deaf and dumb by analyzing the common trends from data and predict the best result. This way, we can handle the situation in a better manner. Deep learning, convolutional neural network, computer vision, Tensorflow etc. modern techniques along with python programming language are being used in this paper to build an application that will detect hand signs and will give accurate and appropriate result. Dataset used in this project is containing static hand gestures images captured with the help of webcam and then preprocessed to feed them to the model making part. The resultant application does live detection using camera feed and give result about what sign means shown by the user. Future scope is also high of this project as it can also be extended to detect even vast multitude of hand gestures, words or even sentences.